The object of the present study was to define the optimal conditions for the quantitative determination of fibronectin in synovial fluid by laser nephelometry and to determine the diagnostic importance of this analysis in rheumatology. The results show that this technique is sensitive (threshold sensitivity is 70 mg/l if the sample is diluted 1/30 and 5 mg/l for a sample diluted to 1/2), accurate, reproducible (coefficient of variation less than 10%), specific, simple and rapid. The response is linear for dilutions from 1/30 to 1/240. Fibronectin P and fibronectin S were determined in 104 subjects who had contracted an arthropathy; 26 cases of rheumatoid polyarthritis, 43 mechanical arthropathies and 35 cases of non-rheumatoid arthritis. the plasma fibronectin does not vary significantly from one group of arthropathies to another but the mean value for the total population of arthropathies is significantly lower than that for a control group. In the synovial fluid the mean value for fibronectin in the patients with rheumatoid arthritis (583 + 76 mg/l) is very considerably higher than that in non-rheumatoid (379 + 58 mg/l) or mechanical (367 + 32 mg/l) arthritis patients. Thus, the determination of synovial fibronectin may provide useful information for the etiological diagnosis of inflammatory arthropathies.